rfPermute (version 2.1.81)

classConfInt: Classification Confidence Intervals

Description

Calculate confidence intervals for Random Forest classifications

Usage

classConfInt(rf, conf.level = 0.95, threshold = 0.8)

Arguments

rf

a randomForest object

conf.level

confidence level for the binom.test confidence interval

threshold

threshold to test observed classification probability against.

Value

A matrix with the following columns for each class and overall:

pct.correct

percent correctly classified

LCI_##, UCI_##

the lower and upper central confidence intervals given conf.level

Pr.gt_##

the probability that the true classification probability is >= threshold

Examples

# NOT RUN {
data(symb.metab)

rf <- randomForest(type ~ ., symb.metab)
classConfInt(rf)

# }